I don't know how to give you more information on this thing, the problem was found also on void linux and a developer has solved it and says he has made it available to everyone but it must be the various Kernel manufacturers that must apply it. I can't tell you anything else, your Kernel developers should do a test with Navi 10 and see the problem in my opinion otherwise we will never get to grips with it if they don't do field tests. I can only tell you that to date 5.6.13 with Navi 10 enabled by the Xorg file it does not work and you have to restart with another Kernel that contains the patch and that has no operating bug ...
I can post you what he wrote in a forum about the resolution of the problem and the link he posted together with the message:
I made the fix, and submitted it to the kernel. AMD then accepted and merged it. Void Linux ships the fix ahead of time (included in 5.6.8 there) but other distros are free to pick it up if they want to. Or eventually, it'll appear in some 5.6 bugfix release, I don't know which. Logged https://voidlinux-ppc.org https://voidlinux-ppc.org/
Le 21/05/2020 à 14:40, Maurizio M a écrit :
I made the fix, and submitted it to the kernel. AMD then accepted and merged it. Void Linux ships the fix ahead of time (included in 5.6.8 there) but other distros are free to pick it up if they want to. Or eventually, it'll appear in some 5.6 bugfix release, I don't know which.
According to;
https://forums.raptorcs.com/index.php/topic,100.msg915.html#msg915
Seems to be:
https://lists.freedesktop.org/archives/amd-gfx/2020-April/049082.html
it has been merged upstream in kernel v5.7-rc5:
59dfb0c64d38 ("drm/amd/display: work around fp code being emitted outside of DC_FP_START/END")
Thanks, Laurent
Le 21/05/2020 à 15:57, Laurent Vivier a écrit :
Le 21/05/2020 à 14:40, Maurizio M a écrit :
I made the fix, and submitted it to the kernel. AMD then accepted and merged it. Void Linux ships the fix ahead of time (included in 5.6.8 there) but other distros are free to pick it up if they want to. Or eventually, it'll appear in some 5.6 bugfix release, I don't know which.
According to;
https://forums.raptorcs.com/index.php/topic,100.msg915.html#msg915
Seems to be:
https://lists.freedesktop.org/archives/amd-gfx/2020-April/049082.html
it has been merged upstream in kernel v5.7-rc5:
59dfb0c64d38 ("drm/amd/display: work around fp code being emitted outside of DC_FP_START/END")
And in linux stable v5.6.13:
6f0d234dad15 ("drm/amd/display: work around fp code being emitted outside of DC_FP_START/END")
Thanks, LAurent
On Thu, 21 May 2020 16:07:59 +0200 Laurent Vivier laurent@vivier.eu wrote:
Le 21/05/2020 à 15:57, Laurent Vivier a écrit :
Le 21/05/2020 à 14:40, Maurizio M a écrit :
I made the fix, and submitted it to the kernel. AMD then accepted and merged it. Void Linux ships the fix ahead of time (included in 5.6.8 there) but other distros are free to pick it up if they want to. Or eventually, it'll appear in some 5.6 bugfix release, I don't know which.
According to;
https://forums.raptorcs.com/index.php/topic,100.msg915.html#msg915
Seems to be:
https://lists.freedesktop.org/archives/amd-gfx/2020-April/049082.html
it has been merged upstream in kernel v5.7-rc5:
59dfb0c64d38 ("drm/amd/display: work around fp code being emitted outside of DC_FP_START/END")
And in linux stable v5.6.13:
6f0d234dad15 ("drm/amd/display: work around fp code being emitted outside of DC_FP_START/END")
and the Void Linux guy confirmed to me a while ago it works OK in the mainline 5.6.13. So Maurizio must have some other issue as well and we need more information.
https://wiki.raptorcs.com/wiki/Enabling_Navi_10_On_Fedora_31 says the fast-reboot feature must be disabled, because the cards can't reinitialize with a different firmware version.
Maurizio, what hardware do you use? Talos, Blackbird, ...? Do you use the AMD card for the skiroot environment (to control petitboot)? Do you have a firmware loaded in BOOTKERNFW partition in the PNOR flash (https://wiki.raptorcs.com/wiki/Add_GPU_Firmware_To_BOOTKERNFW)? What symptoms do have (kernel crash, no output at all, ...)?
Dan